Researchers from the University of Helsinki created a computer technique that uses artificial intelligence (AI) to predict a person's thoughts. The new technique monitors human brain signals to “imagine” what a person is thinking and create that image. With this method, computers can produce new information they have never seen before. The method is called neuroadaptive generative modeling.
Previous computer interfaces have been able to perform one-way communication from brain to computer, but not two way. This is the first study where a computer’s presentation of information and brain signals was modeled simultaneously by an AI program.
The images generated by the computer were evaluated by the participants. They nearly perfectly matched with the features the participants were thinking of. Source: Cognitive Computing Research Group
The team gathered 31 volunteers for the study, and they were tasked to evaluate the method’s effectiveness. The participants were shown hundreds of AI-generated images of a diverse group of people. They were asked to concentrate on certain features on faces shown in a rapid series. While this was happening, the subject’s EEGs were fed into the neural network.
Images generated by the computer were evaluated by the participants. Results showed that the computer system's images were 83% to the images the were thinking of.
Face generation is one way that this technique could be used. The team says that it could be used to help people strengthen their creativity skills by augmenting creativity. It could also be used to gain an understanding of human perception and mind processes.
A paper on this technology was published in Scientific Reports.
